Document not found! Please try again

Photoassisted water-gas shift reaction over platinized titanium dioxide

S. Sato, and J. M. White. J. Am. Chem .... Ryuhei Nakamura and Shinri Sato. The Journal ... Takashi Hisatomi , Kentaro Teramura , Jun Kubota , Kazunar...
0 downloads 0 Views 616KB Size
7206

J . Am. Chem. SOC.1980, 102, 7206-7210

compound and/or pairwise shape descriptor which best fits the high end of the activity data base should probably be chosen. Also, those reference compounds having diverse multiple group substitutions, e.g., X3 # X4 and both # H for the triazines, might be preferable. Such compounds cover greater structural diversity and might extend the utility of a correlation with respect to the possible range of group substitutions. Lastly we conclude this report by pointing out the significance of the work reported here: Three-dimensional molecular shape descriptors have been defined for a set of Baker triazines which are DHFR inhibitors. A maximum of two shape descriptors, and one physicochemical feature, are needed to explain the variance in DHFR inhibition activity. No compounds in the data base had to be eliminated in the QSAR analysis as outlyers. The QSAR constructed with the use of molecular shape descriptors is superior to that of Silipo and Hanschl8 which is based solely upon physicochemical and substructural features. Hence, our study demonstrates a clear instance where the systematic consideration of three-dimensional molecular geometry is essential to describing biological drug potency. Note Added in Proof: On January 15, 1980 part of this work was presented for the first time in a seminar at Warner-Lambert Company, Ann Arbor, MI. This author reported the predicted activity of compound 38 (log (1/C) = 7.61 and 7.72, Table VII). Dr. M. L. Black, of Warner-Lambert, was in attendance and noted that compound 38 had been synthesized by Warner-Lambert but never tested for DHFR activity. On February 6, 1980 a sample

of compound 38 was sent to this author by Dr. Black. I immediately forwarded the sample to Professor C. Hansch at Pomona College for testing. The observed activity according to Professor Hansch and colleagues, based upon the correction for enzyme species as noted in Table VI, is 7.36. A second QSAR study of bovine and rat liver DHFR inhibition by the 3-position substituted 2,4-diaminotriazines, reported in ref 19, is complete and the results are to be published in Arch. Biochem. Biophys. The predicted activity for bovine liver DHFR is 6.58 while the observed value is 6.73 for compound no. 38. These successful predictions give additional support for the reliability of the QSAR reported in this paper and suggest that the QSAR might be used to develop new lead compounds. Acknowledgment. The theory and corresponding computer programs to calculate the shape descriptors were developed under private funding. The application of the shape theory to the Baker triazines were funded under a contract from the National Cancer Institute (Contract No. N01-CP-75927) and a grant from the National Science Foundation (Grant No. ENV77-24061). The author appreciates helpful discussions with Dr. R. D. Cramer, I11 of Smith-Kline Laboratories, Dr. R. D. Battershell of Diamond Shamrock Corporation, Dr. J. Y. Fukunaga of Schering-Plough Corporation, Dr. H. Jabloner of Hercules Incoprated, Professor C. Hansch of Pomona College, and members of A.J.H.'s laboratory during the course of this study. The shaded graphics pictures of the molecular fragments shown in Figure 5 were generated by Dr. Nelson Max of Lawrence Livermore Laboratory from supplied atomic coordinates.

Photoassisted Water-Gas Shift Reaction over Platinized Ti02 Cata1y st s S. Sat0 and J. M. White* Contribution from the Department of Chemistry, University of Texas, Austin, Texas 78712. Received April 24, I980

Abstract: The reaction of gas-phase water with CO (water-gas shift reaction) over platinized, powdered Ti02 is found to be catalytic under UV illumination. The reaction kinetics has been studied at temperatures from 0 to 60 "C. At 25 O C , the reaction is zero order both in CO and H 2 0 when pco 2 0.3 torr and pHlo 1 5 torr and the activation energy is 7.5 kcal/mol.

The wavelength dependence of the reaction rate shows a cut-off near, but slightly below, the band gap of TiO,. The quantum efficiency of the reaction is about 0.5% at 25 'C. A mechanism is proposed which involves the photodecomposition of H 2 0 over platinized TiOz.

1. Introduction Heterogeneous photocatalysis is a relatively new branch of catalysis and in almost all cases semiconductors are used because band-gap photon absorption leads to spatially separated electrons and holes which can be used in oxidation-reduction reactions.' Metallized semiconductors have recently appeared which are more effective in some cases than semiconductors alone. For example, Bulatov and Khideke12have reported the photodecomposition of acidified water by using platinized Ti02 (Pt/TiOz). Kraeutler and Bard3 have reported the selective decomposition of liquid acetic acid to methane over illuminated Pt/Ti02. In the gas phase, Hemminger et aL4 have claimed the photocatalytic production (1) See e.g. M. Formenti and (1979).

S.J. Teichner, Catalysis (London) 2, 87

(2) A. V. Bulatov and M. L. Khidekel, Izu. Akod. Nauk SSSR, Ser. Khim., 1902 (1976). ( 3 ) B. Kraeutler and A. J. Bard, J . Am. Chem. SOC.,100, 2239 (1978).

of methane from COz and gaseous H 2 0 using a SrTi03 crystal contacted with a Pt foil. We have recently begun to investigate the photocatalytic properties of T i 0 2 and Pt/Ti02 and have found that UV-illuminated Pt/Ti02, but not TiOz alone, catalytically decomposes liquid H20.5 Platinized TiOz also photocatalyzes the reactions of gaseous H 2 0 with hydrocarbons,6 active carbon: and lignite' to produce H2 and C02, all of which are thermodynamically unfavorable. We report here another photoassisted reaction over Pt/Ti02, the reaction of gaseous H 2 0 with CO (water-gas shift reaction). Although this reaction is not thermodynamically uphill (4) J. C. Heminger,R. Carr, and G. A. Somorjai, Chem. Phys. Lett., 57, 100 (1978). ( 5 ) S.Sato and J. M. White, Chem. Phys. Lett., in press. (6) S. Sato and J. M. White, Chem. Phys. Lett., 70, 131 (1980). (7) S.Sat0 and J. M. White, Ind. Eng. Chem. Prod. Res. Dm., submitted

for publication.

0002-7863/80/1502-7206$01 .OO/O 0 1980 American Chemical Society

J . Am. Chem. Soc., Vol. 102, No. 24, 1980 7201

Wafer-Gas Shift Reaction over Ti02 Catalysts (AGO = -6.8 kcal/mol) over illuminated Pt/Ti02, we believe it does involve the photodecomposition of water. Thus the present study, in addition to having intrinsic interest, serves to promote to our understanding of thermodynamically unfavorable processes over illuminated Pt/Ti02. The purpose of this article is to demonstrate that the photoassisted water-gas shift reaction can be catalytically accomplished over platinized titania at room temperature and that a mechanism which is compatible with the data involves the photodecomposition of water. It is not possible to establish clearly many of the molecular level details of this reaction. This awaits further experimentation involving the combination of optical and electron spectroscopic techniques. 2. Experimental Section Ti02 (anatase) was obtained from MCB, doped in flowing H2 for 6 h at 700 OC, cooled in H2. and stored in a sample vial. Hydrogen was used after passage through a molecular sieve trap at liquid N2temperature. Platinized Ti02(-2 wt % Pt) was prepared by the photolysis of hexachloroplatinatedissolved in a Na2C03-aceticacid buffer solution.* After photolysis the sample was washed with distilled water until C1could not be detected and then was dried in a desiccator. The BET surface area of Pt/Ti02 was about 11 m2/g. Oxygen-18 H20 (99.3%) was obtained from Prochem. Carbon-13 CO (90%)9 was used after passing through a liquid N2 trap. The apparatus consisted of a vacuum system (base pressure = 1 X lo-' volume), and torr), a gas handling system, a reaction system (180" a gas analysis system, most of which were made of Pyrex glass. The reaction system was equipped with a circulation pump, a quartz reaction cell, and a storage tube of distilled water which was outgassed several times at dry ice temperature. The catalyst (0.25 g) was spread uniformly on the flat bottom of the reaction cell and outgassed for 3 h at 200 OC. When the water-gas shift reaction was studied, CO was introduced into the reaction system and circulated through the water storage tube to supply water vapor. When low H20 pressures were needed, the water storage tube was cooled to a temperature at which the desired H20 vapor pressure was obtained. During reaction the tube was isolated from the reaction system. After the reaction cell temperature was set with a water bath, the catalyst was illuminated with a 200-W high-pressure Hg lamp that was filtered through a quartz cell filled with NiS04 solution to remove heat. The gas mixture was sampled at various times and after passage through a cold trap at about -1 10 OC to remove H 2 0 was analyzed by a mass spectrometer (CEC 21-614). The sensitivity of the mass spectrometer for each gas was calibrated by using a gas mixture of known composition. 3. Results 3.1. Reaction of Gas-Phase H2O with Doped Ti02. The hydrogen-doped Ti02 (without Pt) used in the present experiments reacts with gaseous H 2 0 to produce H2 under UV illumination a t room temperature. This reaction was accompanied by the production of a small and variable amount of C02 which we believe arises from the photodecomposition of adsorbed carbonate. The rate of H2 formation dropped to zero after a 2-h irradiation, and the maximum amount of Hz formed was about 3 X mol. This reaction was not affected by the presence of CO (i.e., the water-gas shift reaction did not take place) but was completely retarded by the addition of 3 X torr of 02.Hydrogen was also produced when the Ti02 was heated in gaseous H 2 0 at temperatures above 200 OC. These results suggest that the H2 formation arises from the activated and noncatalytic reaction of H 2 0with a strongly reduced form of Ti02such as Ti203or TiO.'O 3.2. Reaction of H 2 0 with Pt/Ti02. When Pt/Ti02 was illuminated in the presence of gaseous HzO, H2 and COz were formed with concomitant formation of a small amount of CHI. As shown in Figure 1, the use of H2180instead of ordinary H 2 0 revealed that a part of the oxygen in C 0 2comes from H20. The ratio of (H2+ 2CH4) to (Ci802+ 1/2Ci60180)is nearly 2:1, suggesting that H20 is photodecomposed on Pt/TiOz and that the oxygen formed reacts with adsorbed carbon and CO. These B. Kraeutler and A. J. Bard, J . Am. Chem. SOC.,100, 4317 (1978). (9) We thank Dr. T. N. Taylor of Los Alamos Scientific Lab for a sample (8)

of ~ 0 . (10) D. R. Kennedy, M. Ritchie, and J. Mackenzie, Trans. Faraday SOC., 54, 119 (1958).

1.05

0

0

40

120

80

Timehin) Figure 1. Products of the reaction of H2'*0 with Pt/Ti02 under UV illumination ( p ~ p c o 24 torr). 0.6

1

E 04 W

2 W E v)

02

20

40

60

80

100

120

140

TIME ( m i d

Figure 2. Time course of the water-gas shift reaction over illuminated Pt/Ti02 at 25 OC (PH~O 24 torr).

carbon-containing species are probably present as the result of acetic acid decomposition during the preparation of Pt/Ti02. The stoichiometric reaction of HzO with doped Ti02, described in section 3.1, should not contribute to these results since it will have gone to completion during the catalyst preparation. Oxygen atom exchange between H 2 0 and C 0 2 present originally on Ti02 may contribute a small amount to the observed Ci60l8O. Hydrogenation of adsorbed carbon will account for the formation of CHI. From Figure 1 the maximum yield of H2 is certainly larger than 0.09 torr (8.7 X mol). This is a factor of 3 larger than the maximum H2yield found for Ti02 and suggests that the addition of Pt increases the activity for H2 production perhaps through a reaction pathway involving the decomposition of water. If the photodecomposition of H 2 0 is indeed involved, Hz should be continuously formed by the addition of some material such as CO, which has a strong affinity for oxygen and forms a nonreactive product such as C02. On this basis we tested for, and found, catalytic activity for the water-gas shift reaction on Pt/TiO,. 3.3. Water-Gas Shift Reaction. General Features. Figure 2 shows a typical time evolution of the water-gas shift reaction over UV-illuminated Pt/Ti02 at 25 OC. Carbon-13 CO (0.6 torr total, 0.54 torr l3C0) was used in order to discriminate against the C02 yield that occurs in the absence of CO(g) (see Figure 1). The water pressure was 24 torr. When the isotopic purity of CO (90% 13C)is taken into account, the stoichiometry of the shift reaction is well satisfied: - A p a = Appc~,= b2. In the dark, the reaction rate was very slow at room temperature and, at 60 OC, its contribution was only 0.6% of the photoprocess. According to Figure 2, the production of Hz proceeds with no marked loss of activity until the carbon monoxide is consumed. Once the CO(g) is gone, however, the activity drops sharply to zero. In fact the H2 pressure passes through a maximum, indicating that some H2is consumed

Sato and White

7208 J . Am. Chem. SOC.,Vol. 102, No. 24, 1980

i

0.2

60

120

180

240

TIMElmin)

Figure 3. Time course of the photoassisted water-gas shift reaction over Pt/TiO, at 0 OC. Plots of COz are omitted as they almost overlap on the plots of H2 ( P H s N 5 torr).

in the latter stages of the experiment summarized in Figure 2. In this case the maximum amount of H2 is 5.8 X 10" mol (1.3 x 1018molecules m-2 of catalyst). As shown in Figure 2, a small amount of 02(g) appears as the pressure of CO drops below 0.06 torr. This pressure gradually increases at the end of reaction and then abruptly jumps to about torr when the CO pressure drops below 0.01 torr. No other products were detected in this experiment. The same was true for other reaction temperatures between 0 and 60 O C . Reproducibility. The photocatalytic activity was reproducible in a series of runs when the reaction was stopped and the system evacuated before the CO pressure dropped below about 0.1 torr. If the reaction proceeded to completion, as in Figure 2, an activity increase was sometimes observed upon repeating the reaction after evacuation for only a few minutes. Outgassing for extended periods, even at room temperature, restored the activity to a reproducible value. In our experiments reproducible and stable activity was achieved after any experiment by outgassing at 150 O C for 1 h. Pressure Dependence at 25 O C . Figure 2 shows that the rate of the photoassisted water-gas shift reaction is almost constant until the pressure of CO falls to about 0.3 torr. Over this range there is a zero-order dependence of the rate on CO pressure. Since the pressure of H 2 0 (24 torr) is much higher than the CO pressure (0.6 torr), the effect of its change during the reaction is negligible. The reaction accelerates slightly as the CO pressure drops below 0.3 torr, indicating a slightly negative order in pco between 0.3 and 0.1 torr. Near completion the rate slows as the CO pressure drops, implying a positive order in pco. These qualitative dependences were confirmed in a series of experiments utilizing initial CO pressure from 0.1 to 0.6 torr. A slight increase in the initial rate (10-20%) was observed when the H 2 0 pressure was reduced from 24 torr to about 5 torr while the initial CO pressure was fixed at about 0.6 torr. The reaction is, therefore, almost zero order in H 2 0 pressure. Temperature Dependence. Figure 3 shows the time dependence of the reaction at 0 OC. The reaction rate is nearly constant over the entire pressure range observed though a slight decline is observed as the CO pressure drops below 0.25 torr. The consumption of CO exceeds slightly the formation of H2 probably due to CO adsorption on the catalyst during the reaction. The C 0 2 production rate (not shown) equals the H2 production rate. At 60 O C (Figure 4) the time evolution is qualitatively like that at 25 "C: an initial constant rate followed by an acceleration at intermediate CO pressures and a retardation at the end of the reaction. However, the CO pressure at which the acceleration begins is higher than that in the reaction at 25 O C . The decline of the rate is not due to a loss of the photocatalytic activity of Pt/Ti02 because the initial rate was reproduced in the next run. When the initial pressure of CO was reduced to 0.5 torr, the rate

20

IO

300

33

50

40

60

70

TIME (min)

Figure 4. Time course of the photoassisted water-gas shift reaction over Pt/Ti02 at 60 OC. Plots of C 0 2 almost overlap on those of H2 @ H ~ O = 24 torr). 20

\I 30

3.2 IO'/T

34

36

38

(K-'I

Figure 5. Arrhenius plot of the rates of photoassisted water-gas shift reaction over Pt/TiO, [pCo = 0.5-0.6 torr, p H P = 24 torr (-5 torr at 0 "C)]. IO0

"I; 200 300 400 CUT-OFF WAVELENGTH (nm)

Figure 6. Effects of cutoff filters on the rate of photoassisted water-gas shift reaction over Pt/TiO, at 25 and 60 O C (pHZoN 24 torr, pCo = 0.5-0.6 torr).

accelerated earlier than in Figure 4 and the retardation period became shorter. The reaction course at 40 O C showed a pattern intermediate between those at 25 and 60 O C . Figure 5 shows an Arrhenius plot of the reaction rates for temperatures from 0 to 60 OC in the region where the reaction takes place a t a constant rate. The activation energy is 7.5 kcal/mol, which is about the same as that estimated for the photoassisted decomposition of liquid water ( N 5 kcal Wavelength Dependence. The wavelength dependence was qualitatively measured by using three cutoff filters: a commercial UV cutoff filter (415-nm cutoff), a Plexiglass filter (3 mm thick, about 380-nm cutoff), and a Pyrex glass filter (3 mm thick, about 275-nm cutoff). As seen from Figure 6, the 415-nm cutoff filter completely retarded the reaction while the Plexiglass and Pyrex filters suppressed the rate to 4 and 6796, respectively. There was no difference in the wavelength dependence at 25 and 60 O C .

J. Am. Chem. Sot., Vol. 102, No. 24, 1980 7209

Water-Gas Shut Reaction over TiOz Catalysts

0.15

IO

20

X,

40

TIME

(min)

50

60

70

Figure 7. Formation of O2 during the photoassisted water-gas shift reaction over Pt/Ti02 a t 25 OC ( p H p N 24 torr).

Formation of 02.The formation of O2during the water-gas shift reaction is surprising in the sense that it is not favored thermodynamically. The fact that it is formed may help in understanding the mechanism of the photoassisted water-gas shift reaction over Pt/Ti02. Typical pressure vs. time curves for CO, H2, C02, and O2are shown in Figure 7. The amount of O2 is very small (less than 5 X lo4 torr) and almost constant in the region where the reaction takes place at a constant rate. After the shift reaction accelerates, O2gradually increases and then maximizes rather abruptly as CO drops below 0.01 torr. At the same time, the H 2 / C 0 2 ratio deviates sharply from unity and the excess of H2over C 0 2 is about twice the amount of O2formed, suggesting the decomposition of H20. The back-reaction occurs readily over Pt, and Figure 7 suggests that the rate of this process increases when the CO pressure drops below lo-' torr. The H2consumption rate observed here is much lower than the H2oxidation rate on a Pt/Ti02 surface not exposed to C0.5 As shown in Figure 7, the rate becomes a little faster after the light is turned off and turning the light source back on produces small amounts of O2and H2 This O2formation pattern was reproducible although the quantitative amounts of O2produced in each region are sensitive to the activity of Pt/Ti02, the reaction temperature and the initial CO pressure. The maximum amount of O2formed (t N 25 min in Figure 7) is not exactly reproduced but tends to decrease as the reaction temperature increases, as the activity decreases or as the initial CO pressure is raised. Reparation Method Dependence. The Pt/Ti02 prepared from undoped T i 0 2 (no H2 treatment) showed much lower (=10%) photocatalytic activity than the Pt/H2-doped T i 0 2 for the water-gas shift and the water decomposition reaction. The activity depends upon the conditions of the H2 doping of Ti02 (temperature, time, and a flow rate of H2). A strongly reduced Ti02 leads to higher activity but an optimum condition was not established. As doping proceeds, Ti02 changes color from white to light blue, indicating the production of a reduced form of Ti02. Preliminary ESCA spectra of the doped and undoped forms revealed no detectable differences. The Pt loading may also affect the photocatalytic activity, but this was not examined in the present study.

4. Discussion Our experiments show that the photoassisted water-gas shift reaction over Pt/Ti02 is catalytic when the catalyst is irradiated with band-gap radiation. The reaction may involve photoinduced electronic processes similar to those noted in photoelectrochemical (PEC) cells' or photoelectrochemical diodes.I2 At the Pt-Ti02 junctions formed by deposition of Pt, the valence and conduction bands will be bent upwards, because charge transfer will occur to equalize the Fermi levels of these materials. The extent to which this occurs for our catalyst cannot be readily determined because the doping level of Ti02 is unknown, because the electronic coupling between Ti02 and Pt is not known and because the Pt particles are small and not fully characterizable by bulk parameters. Upon irradiation and excitation of an electron (1 1) See e.g.: H. P. Maruska and A. K. Ghosh, Sol. Energy, 20, 443 (1978). (12) A. J. Nozik, Appl. Phys. Letr., 30, 567 (1977).

Figure 8. Pictorial Model of the photoinduced redox processes occurring at the Pt/Ti02 surface.

into the conduction band of Ti02, charge separation will compete with hole-electron recombination. Holes will tend to migrate to the Pt-TiO, interface, and the electrons will migrate away from it. When a photostationary state is reached, the bending of the valence and conduction bands in the semiconductor will be reduced and the electrons and holes can be characterized by quasi-Fermi levels." As a result of these photovoltage effects, we expect (1) that recombination of electrons and holes will become relatively more important than would be expected on the basis of the "dark" energy level diagram, (2) that holes will be found at both the Pt/Ti02 and the Ti02/gas interfaces, and (3) that electrons will be captured at the Pt particles. If so, the photocatalyzed processes we have observed here can be described by using the diagram (analogous to a shorted electrochemical cellI3) shown in Figure 8. In this mechanism, holes come to the Ti02/gas interface where they react with adsorbed water to form an adsorbed OH radical and a proton. The latter diffuses to the Pt, probably mediated by the physisorbed water that is present, where it is reduced. When CO is present, it reacts with the O H radical to give CO2 and a hydrogen atom which recombines with other hydrogen atoms at Pt sites. In the absence of CO, the adsorbed O H radical will either react with a second hole to form H+ and a chemisorbed oxygen atom or two O H radicals will recombine to form H202. Gas-phase oxygen would then be formed either by recombination of oxygen atoms or by the rapid decomposition of H202. The back-reaction, H2(g) 1/202(g) H20(g), occurs readily on Pt14and even occurs when liquid water is present,s but at a slower rate. Consequently, in the absence of C O no measurable 02 is evolved. In this case the only way a net amount of H2 is evolved is to use surface carbon, present on'the catalysts as a result of preparative techniques, to scavenge oxygen, and to make C02 (see Figure 1). The above descriptive mechanism is also applicable to the photolysis of liquid-phase water on Pt/Ti0,5 where the photodissociation rate is more competitive with the back-reaction. There is, however, a problem in view of the fact that the Pt/Ti02 PEC cell requires an external potential for water d e c o m p ~ s i t i o n . ~ ~ ~ ' ~ The reason for this requirement is not obvious. There are conflicting views regarding the position of the flat band potential of Ti02 with respect to the H+/H2 redox potential. It has been reported as 50-mV positive" and 100-mV negative.16 In the former case, electrons, arriving at the Pt electrode of a PEC cell must pass an energy barrier to reduce protons to hydrogen (this assumes the pressure is 1 atm). In the latter case, the reaction should occur under short circuit conditions. In either case, overvoltages and other voltage drops will contribute to the inefficiency and to the external potential requirements. Our results, including those for liquid water,5 show that this reaction does occur without an external potential when the solid photoactive material involves small Pt particles dispersed on powdered Ti02. One of the more important differences between our experimental conditions and those in PEC cells is the pressure. The total pressure in our experiments lies between 20 and 25 torr,

+

-

(13) A. J. Bard, J . Phorochem. 10, 59 (1979). (14) M. Boudart, D. M. Collins, F. V. Hanson, and W. E. Spicer, J . VUC. Sci. Technol., 14, 441 (1977). (15) M. S . Wrighton, D. S. Ginley, P. T. Wolczanski, A. B. Ellis, D. L. Morse, and A. Linz, Proc. Natl. Acad. Sci. U.S.A. 72, 1518 (1975). (16) M. Tomkiewicz, J . Electrochem. Soc., 126, 1505 (1979).

7210 J . Am. Chem. SOC.,Vol. 102, No. 24, 1980

Sato and White

and the hydrogen pressure never exceeds 0.7 torr. The H+/H2 redox potential under these conditions will certainly become more positive with respect to the Ti02 flat band potential and will allow the reaction to proceed more efficiently. Other differences must also be considered: (1) There will be more surface defects on powdered Ti02 than on crystalline samples; the attending defect electronic states may be important. (2) The small Pt particles present on our catalyst may have electronic properties different from bulk Pt. (3) Unlike PEC cells, potentially ~ignificant'~ direct chemical bonding between Pt and Ti02may be present here. The importance of each of these cannot be established on the basis of our data. Another significant factor in the activity of the solid phase is the reduction of the Ti02 in the preparation procedure. As noted by Wrighton et al.,I5 this is important in photoelectrochemical cells. One reason is the lower resistivity of the H,-doped T i 0 2 which enhances photoconductivity. We expect this same change in conductivity to be important in our powdered catalysts. Changes in electronic band structure upon creation of oxygen vacancies are also expected, but this can not be examined by kinetic methods. Enhancement of the adsorption of reacting species on the reduced T i 0 2 surfaces does not appear to be important. Focusing now on the water-gas shift reaction, we note that the photodriven dissociation of water is indeed implicated since O2 evolution is noted (Figures 2, 4, and 7). With the assumption of such a model, the reaction steps probably include (1)-(7). The

-+ -

+ H20 p + + OH(a) p+

CO(a)

+ OH(a) H+

e-

+ H+ O(a) + H+

OH(a)

(1) (2) (3)

COz(g) + H(a) H(a)

(4)

(6)

2H(a) H2(g) (7) intermediates are either denoted in Figure 8 or described in the text that accompanies it. It is not clear whether Pt participates in steps 3,4, and 5 since it may be covered with H(a) during the reaction, as observed in the hydrogen electrode reaction over Pt dispersed on graphite.I8 The time course of the water-gas shift reaction consists of three regions: (1) region 1 where the pressure of CO is relatively high (17) S. J. Tanster, S. C. Fung, and R. L. Garten, J. Am. Chem. Soc., 100, l(1978). (18) V.S.Bagostzky, L. S. Kanevsky, and V. Sh. Palanker, Electrochim. Acta, 18, 473 (1973);A. Katayama and H. Kita, J. Res. Inst. Catal., Hokkaido Uniu., 26, 131 (1978).

and the reaction occurs at a constant rate, (2) region 2 where the reaction gradually accelerates with time, and (3) region 3 where the depression of the rate and the formation of O2are observed (Figures 2, 4, and 7). When the CO pressure is high (region l), the CO coverage on the catalyst surface would be saturated so that the reaction rate is dependent of the C O pressure. We believe this coverage on Pt inhibits H2 evolution as observed in the reaction of water with active carbonI9 and lignite7 over illuminated Pt/TiOz. The coverage of CO on Pt begins to decrease when the CO pressure is reduced below a certain temperaturedependent value. As a result, the reaction rate accelerates (region 2). When the C O pressure is low (region 3), the diffusion of C O through the layer of physisorbed water on TiOz may limit the rate (step 3). As a result the overall reaction rate is suppressed and the coverage of O(a) may increase so that O2is produced. Some strongly adsorbed CO would however remain on Pt even after the gaseous CO is consumed and tend to inhibit the reaction of O2 with Hz. This adsorbed CO may be slowly removed and allow the back-reaction to occur more rapidly (Figure 7). When the reaction temperature is increased, the amount of CO adsorbed on Pt will drop for a given CO pressure. Accordingly, acceleration of the shift reaction starts at higher C O pressures as indicated in Figures 2-4. The migration of H(a) and O(a) on the catalyst surface, which results in their reaction to form HzO, may become important at the end of the reaction at 60 O C (Figure 4). The reaction order with respect to CO pressure is thus explained in terms of the inhibition by CO (zero and negative order) and the CO diffusion (positive order). The zero-order dependence of the rate on H 2 0 pressure is reasonable since the catalyst surface would be covered with physisorbed H20even at relatively low H 2 0 pressures. The wavelength dependence of the rate is similar to the wavelength response of the photocurrent in a Pt/W-doped T i 0 2 PEC cell studied by Wrighton et al.I5 According to their results, a TiOz electrode shows a photocurrent threshold near 400 nm and a saturation near 350 nm, while the onset 'of W-doped TiOz is shifted to shorter wavelengths by nearly 50 nm; Hz-doping and -platinizing T i 0 2 may have a similar effect. The quantum efficiency of the present reaction is low (about 0.5% in region 1 a t room temperature). This may be improved by changing the level of H2 doping and the Pt loading.

Acknowledgment. We thank Mr. B. E. Koel for ESCA analysis of the T i 0 2 samples and Mr. S.-K. Shi for his help in the preparation of our experiments. This work was supported in part by the Office of Naval Research. (19) S.Sato and J. M. White, unpublished result.